Title: Design and Evaluation of an Intelligent Web Application for At-Home Fitness and Personal Health Tracking
Authors: Pham Hoang Quan Tran Quang Truong
Volume: 9
Issue: 10
Pages: 152-162
Publication Date: 2025/10/28
Abstract:
The consequences during the COVID-19 pandemic were not only due to the virus itself. As a consequence of 'social distancing' regulations and subsequent restrictions implemented during the early phase of the pandemic, the healthcare system, and the general population's physical and mental health suffered greatly. With difficulty in access caused by social distancing regulations, people did not engage in physical activities and were not able to attend healthcare facilities like gyms and yoga studios. In response to this, the proposed study developed an AI-based web solution designed to provide safe and effective guided home exercise to customers. Here, the use of advanced technology that incorporates deep learning and computer vision is described. Particularly, the system employs Google's MediaPipe Pose model to "watch" and "follow" users, while a yoga poses classification model developed in Keras assesses and modifies a user's yoga pose in real-time. The system provides feedback, counts repetitions, and detects exercise poses to ensure users maintain correct form. In addition to guided exercise, the system offers electronically supported health maintenance through blink detection and fatigue monitoring, alerting users to take breaks during extended computer use, as well as working through interactive, movement-based games and other training activities. The website features were developed using Flask (Python) for the backend, with the frontend constructed using HTML, CSS, and Bootstrap. This configuration results in a lightweight system that is easy to configure and adaptable to numerous interfaces. According to experimental findings, this system successfully identifies the location of 32 joints within the human anatomy, records the frequency of a user's workout sessions, and differentiates various yoga positions. The study points to the potential applications of AI in the field of fitness and personal healthcare, especially after the pandemic, when the demand for smart, safe, and easy-to-use remote workout options has increased.